xml handling in shell scripts

[DamienIrving]

Most of the scripts in the old cct repo have the following code block for dealing with a cdscan xml file:

# Check if input is an XML file
inbase=`basename $infile`
extn=`expr match "${inbase}" '.*\.\(.*\)'`
if [ $extn = 'xml' ] ; then
  tmp_in=$TMPDIR/xml_concat.$$.nc
  python ${CWSL_CTOOLS}/utils/xml_to_nc.py None $infile $tmp_in
  infile=$tmp_in
fi

As you can see I've inserted ${CWSL_CTOOLS} to make it compatible with the CWSL plugin, but I'm wondering what to do with $TMPDIR? The nino34.sh script uses $temp_dir which is defined as follows:

temp_dir=$(mktemp -d)
function cleanup {
    rm -rf $temp_dir
}
trap cleanup EXIT

Should I do the same?

[captainceramic]

I like that method of using a temp directory and then trapping the exit to remove it
As for the XML checking, I don't really like it as I would prefer not to use the xml catalogues, but if we must I think that method works.
The only change I would make is a switch from using backticks to using the $() - i.e. using:

inbase=$(basename $infile)

instead of

inbase=`basename $infile`